Energy consumption is a vital concern when implementing distributed decision fusion in most wireless sensor networks. This paper studies the impact of sensor censoring on the decision fusion performance when the number of sensors is unknown at the fusion center. The global decision rule adopted at the fusion center is the Chair-Varshney fusion rule modified to take account of the unknown network size. It is shown that under the assumption of equally likely hypotheses, allowing more transmitting sensors does not necessarily yield better decision fusion; rather, there exists a censoring probability threshold below which the increase in the number of active sensors just incurs more intra-network communication overhead but will not improve the global decision performance. Our findings establish that the design of energy-efficient local detection rules should commence with the censoring rate threshold.